mirror of
https://github.com/yuzu-emu/yuzu.git
synced 2024-08-25 06:27:18 +02:00
gl_rasterizer: Use NV_vertex_buffer_unified_memory for vertex buffer robustness
Switch games are allowed to bind less data than what they use in a vertex buffer, the expected behavior here is that these values are read as zero. At the moment of writing this only D3D12, OpenGL and NVN through NV_vertex_buffer_unified_memory support vertex buffer with a size limit. In theory this could be emulated on Vulkan creating a new VkBuffer for each (handle, offset, length) tuple and binding the expected data to it. This is likely going to be slow and memory expensive when used on the vertex buffer and we have to do it on all draws because we can't know without analyzing indices when a game is going to read vertex data out of bounds. This is not a problem on OpenGL's BufferAddressRangeNV because it takes a length parameter, unlike Vulkan's CmdBindVertexBuffers that only takes buffers and offsets (the length is implicit in VkBuffer). It isn't a problem on D3D12 either, because D3D12_VERTEX_BUFFER_VIEW on IASetVertexBuffers takes SizeInBytes as a parameter (although I am not familiar with robustness on D3D12). Currently this only implements buffer ranges for vertex buffers, although indices can also be affected. A KHR_robustness profile is not created, but Nvidia's driver reads out of bound vertex data as zero anyway, this might have to be changed in the future. - Fixes SMO random triangles when capturing an enemy, getting hit, or looking at the environment on certain maps.
This commit is contained in:
parent
32485917ba
commit
41a4090320
@ -61,7 +61,8 @@ constexpr std::size_t NUM_CONST_BUFFERS_BYTES_PER_STAGE =
|
||||
constexpr std::size_t TOTAL_CONST_BUFFER_BYTES =
|
||||
NUM_CONST_BUFFERS_BYTES_PER_STAGE * Maxwell::MaxShaderStage;
|
||||
|
||||
constexpr std::size_t NumSupportedVertexAttributes = 16;
|
||||
constexpr std::size_t NUM_SUPPORTED_VERTEX_ATTRIBUTES = 16;
|
||||
constexpr std::size_t NUM_SUPPORTED_VERTEX_BINDINGS = 16;
|
||||
|
||||
template <typename Engine, typename Entry>
|
||||
Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry& entry,
|
||||
@ -193,7 +194,7 @@ void RasterizerOpenGL::SetupVertexFormat() {
|
||||
// avoid OpenGL errors.
|
||||
// TODO(Subv): Analyze the shader to identify which attributes are actually used and don't
|
||||
// assume every shader uses them all.
|
||||
for (std::size_t index = 0; index < NumSupportedVertexAttributes; ++index) {
|
||||
for (std::size_t index = 0; index < NUM_SUPPORTED_VERTEX_ATTRIBUTES; ++index) {
|
||||
if (!flags[Dirty::VertexFormat0 + index]) {
|
||||
continue;
|
||||
}
|
||||
@ -231,9 +232,11 @@ void RasterizerOpenGL::SetupVertexBuffer() {
|
||||
|
||||
MICROPROFILE_SCOPE(OpenGL_VB);
|
||||
|
||||
const bool use_unified_memory = device.HasVertexBufferUnifiedMemory();
|
||||
|
||||
// Upload all guest vertex arrays sequentially to our buffer
|
||||
const auto& regs = gpu.regs;
|
||||
for (std::size_t index = 0; index < Maxwell::NumVertexArrays; ++index) {
|
||||
for (std::size_t index = 0; index < NUM_SUPPORTED_VERTEX_BINDINGS; ++index) {
|
||||
if (!flags[Dirty::VertexBuffer0 + index]) {
|
||||
continue;
|
||||
}
|
||||
@ -246,16 +249,25 @@ void RasterizerOpenGL::SetupVertexBuffer() {
|
||||
|
||||
const GPUVAddr start = vertex_array.StartAddress();
|
||||
const GPUVAddr end = regs.vertex_array_limit[index].LimitAddress();
|
||||
|
||||
ASSERT(end >= start);
|
||||
|
||||
const GLuint gl_index = static_cast<GLuint>(index);
|
||||
const u64 size = end - start;
|
||||
if (size == 0) {
|
||||
glBindVertexBuffer(static_cast<GLuint>(index), 0, 0, vertex_array.stride);
|
||||
glBindVertexBuffer(gl_index, 0, 0, vertex_array.stride);
|
||||
if (use_unified_memory) {
|
||||
glBufferAddressRangeNV(GL_VERTEX_ATTRIB_ARRAY_ADDRESS_NV, gl_index, 0, 0);
|
||||
}
|
||||
continue;
|
||||
}
|
||||
const auto info = buffer_cache.UploadMemory(start, size);
|
||||
glBindVertexBuffer(static_cast<GLuint>(index), info.handle, info.offset,
|
||||
vertex_array.stride);
|
||||
if (use_unified_memory) {
|
||||
glBindVertexBuffer(gl_index, 0, 0, vertex_array.stride);
|
||||
glBufferAddressRangeNV(GL_VERTEX_ATTRIB_ARRAY_ADDRESS_NV, gl_index,
|
||||
info.address + info.offset, size);
|
||||
} else {
|
||||
glBindVertexBuffer(gl_index, info.handle, info.offset, vertex_array.stride);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
@ -268,7 +280,7 @@ void RasterizerOpenGL::SetupVertexInstances() {
|
||||
flags[Dirty::VertexInstances] = false;
|
||||
|
||||
const auto& regs = gpu.regs;
|
||||
for (std::size_t index = 0; index < NumSupportedVertexAttributes; ++index) {
|
||||
for (std::size_t index = 0; index < NUM_SUPPORTED_VERTEX_ATTRIBUTES; ++index) {
|
||||
if (!flags[Dirty::VertexInstance0 + index]) {
|
||||
continue;
|
||||
}
|
||||
|
@ -488,6 +488,15 @@ void RendererOpenGL::InitOpenGLObjects() {
|
||||
|
||||
// Clear screen to black
|
||||
LoadColorToActiveGLTexture(0, 0, 0, 0, screen_info.texture);
|
||||
|
||||
// Enable unified vertex attributes and query vertex buffer address when the driver supports it
|
||||
if (device.HasVertexBufferUnifiedMemory()) {
|
||||
glEnableClientState(GL_VERTEX_ATTRIB_ARRAY_UNIFIED_NV);
|
||||
|
||||
glMakeNamedBufferResidentNV(vertex_buffer.handle, GL_READ_ONLY);
|
||||
glGetNamedBufferParameterui64vNV(vertex_buffer.handle, GL_BUFFER_GPU_ADDRESS_NV,
|
||||
&vertex_buffer_address);
|
||||
}
|
||||
}
|
||||
|
||||
void RendererOpenGL::AddTelemetryFields() {
|
||||
@ -656,7 +665,13 @@ void RendererOpenGL::DrawScreen(const Layout::FramebufferLayout& layout) {
|
||||
offsetof(ScreenRectVertex, tex_coord));
|
||||
glVertexAttribBinding(PositionLocation, 0);
|
||||
glVertexAttribBinding(TexCoordLocation, 0);
|
||||
glBindVertexBuffer(0, vertex_buffer.handle, 0, sizeof(ScreenRectVertex));
|
||||
if (device.HasVertexBufferUnifiedMemory()) {
|
||||
glBindVertexBuffer(0, 0, 0, sizeof(ScreenRectVertex));
|
||||
glBufferAddressRangeNV(GL_VERTEX_ATTRIB_ARRAY_ADDRESS_NV, 0, vertex_buffer_address,
|
||||
sizeof(vertices));
|
||||
} else {
|
||||
glBindVertexBuffer(0, vertex_buffer.handle, 0, sizeof(ScreenRectVertex));
|
||||
}
|
||||
|
||||
glBindTextureUnit(0, screen_info.display_texture);
|
||||
glBindSampler(0, 0);
|
||||
|
@ -107,6 +107,9 @@ private:
|
||||
OGLPipeline pipeline;
|
||||
OGLFramebuffer screenshot_framebuffer;
|
||||
|
||||
// GPU address of the vertex buffer
|
||||
GLuint64EXT vertex_buffer_address = 0;
|
||||
|
||||
/// Display information for Switch screen
|
||||
ScreenInfo screen_info;
|
||||
|
||||
|
Loading…
Reference in New Issue
Block a user